英语原文共 8 页，剩余内容已隐藏，支付完成后下载完整资料
毕业设计(论文)
外文资料翻译
题 目： 工件表面缺陷的图像识别技术及应用
说明：
封面之后放英文原文；英文原文之后放译文。
原文：
Defect detection method for complex surface based on human visual characteristics and feature extracting
ABSTRACT
Aimed at the problem of strong background interference introduced in digital image processing from complex surfaces under industrial defect detection, a method for complex surface defect detection based on human visual characteristics and feature extracting is proposed. Inspired by the visual attention mechanism, defect areas can be identified from the background noise conveniently by human eyes. We introduce the improved grayscale adjustment and frequency-tuned saliency algorithm combined with the salient region mask obtained by dilation and differential operation to eliminate the background noise and extract defect areas. Meanwhile the directional feature matching and merging algorithm is applied to enhance directional features and retain details of defects. Testing images are captured by our established detecting system. Experimental results show that our method can retain defect information completely and achieve considerable extracting efficiency and detecting accuracy.
Keywords: machine vision, complex surface, defect detection, human visual characteristics, feature extracting, image matching and merging
- INTRODUCTION
Industrial inspection of surface quality through digital image processing has been developed and widely applied in many fields with high accuracy and fast speed, such as the quality detection of large aperture fine optical elements1 , the surface detection of ceramic elements2 and leather3 and the crack detection on concrete surface4 . However, for products with complex surfaces or irregular shapes such as laptops and other digital devices, it can be quite difficult to obtain high- quality images for inspection directly because areas with defects are easily submerged by enormous noise in the background.
Detecting methods for complex surfaces have been proposed in recent years including using multi-angle illumination5 and artificial neural network training6 to improve the detecting accuracy. Nevertheless, the methods mentioned above require refined illumination system adjustment and strong computing capability which often cannot be satisfied in complicated industrial environment.
In this paper, we introduce a detecting method for complex surface defects based on human visual characteristics and directional feature extracting. The defect areas are acquired combining the human visual saliency model with the salient region mask generated by differential dilation. The directional features of defects can be extracted by image matching and merging to retain defect information. Experimental results indicate that our method can extract sufficient information of defects and achieve considerable detecting accuracy.
2. IMPROVED GRAYSCALE ADJUSTMENT AND FREQUENCY-TUNED SALIENCY ALGORITHM
2.1 Typical image analysis
2.1.1 Illumination analysis
The illuminating distribution on testing surfaces is closely related to surface shapes when using optical imaging detecting methods. Ideally, testing surfaces can receive uniform illumination which contributes to improving the quality and
efficiency of image processing. In industrial detection however, surfaces under testing are often complex in shape which leads to uneven illumination. Figure 1 a) shows an original image sample from a laptop and Figure 1 b) shows the grayscale distribution of the sample with significant uneven illumination due to the irregular appearance. In addition, the roughness of the surface also affects the illuminating distribution in local areas shown in Figure 1 c). Therefore, it is necessary to reduce the effect of uneven illumination and roughness on image processing in complex surface defect
Figure 1. a) The original image containing surface defects b) the overall grayscale distribution with significant variation c) the grayscale distribution in local areas with fluctuation.
2.1.2 Feature analysis of defects
Surface defects vary in scale, shape and position with no specific rules. As is shown in Figure 2, the type of defects includes dot, segment and other types with larger areas such as line and block. The defects can cover either a large scale across the surface or gather in local areas with different contrast to the background. At the pixel level, defect regions consist of a series of pixels with similar grayscale value. The pixels values are obviously different from the grayscale value of the background and attract the human eyersquo;s attention easily.
Figure 2. Images of typical surface defects a) dot b) segment c) line d) block. The line and block type can be regarded as a set of dots and segments.
2.2 Improved grayscale adjustment and frequency-tuned saliency algorithm
Uneven spatial grayscale distribution of the image distracts the human eyersquo;s attention from defect areas. The improved grayscale adjustment and frequency-tuned saliency algorithm (IGAFTS) is proposed to adjust grayscale change of the image and enhance the grayscale contrast of defect areas. As the flow chart shown in Figure 3, the original image is segmented to separate the spatial gra
剩余内容已隐藏，支付完成后下载完整资料
资料编号：[254059]，资料为PDF文档或Word文档，PDF文档可免费转换为Word
以上是毕业论文外文翻译，课题毕业论文、任务书、文献综述、开题报告、程序设计、图纸设计等资料可联系客服协助查找。
